We are interested in how complex behaviors, learned and otherwise, are generated by the brain. To study this, we use rats, experimentally tractable models capable of mastering a variety of motor and cognitive tasks. To help us decipher how the underlying neural circuits function, we have developed a unique experimental infreastructure, including a fully automated high-throughput training system that can be combined with continuous long-term neural recordings and high-resolution 3D behavioral tracking. We further augment our experimental platform with a variety of circuit dissection tools (optogenetics, pharmacogenetics etc) and sophistictaed analytical and computational approaches. In collaboration with DeepMind, we also developed a virtual rodent - a biomechanically realistic model of a rat that is controlled by an artifical neural network and can recapitulate many of the behaviors of real rats. The ultimate goal of our research is to deliver a mechanistic description of how the mammalian brain controls complex behaviors and address how the underlying neural circuits are compromised in disease, including autism.